Hi! Thank you very much for making this overview! There are a couple of things we'd like to point out: 1) You didn't setup the camera, and proceeded with a default one. That's why the model looks a little bit distorted - the shots were certainly made with a wide-angle lens, while you've been facebuilding with 50mm on a full-frame sensor. In case when you don't know camera settings automatic focal length estimation (checkbox in camera settings) can do a pretty good job! 2) The actor's facial expression is not neutral, that's OK if the expression is not too dramatic and so you can use your imagination to fit the face, but you have to know what you're doing very well. ;) 3) You start with nose and eye corners, which is a recommended way, but then you instantly switch to the face edges - the jawline, the chin, etc, while there are some better features position of which you can easily find and which will give you better results - like the corners of the mouth, ears, eyebrows (which are displaced here due to the tensed face. 🤘
Thanks for the tips! (And for the great tools!) 1) I didn't know about the automatic focal length estimation, that great! (Although, I did shoot this footage with a 50mm lens on a Canon 5D.) I'll definitely be experimenting with this. 2) Yep, he's not in a neutral face position, but at work, when we get shots in, the actors never have neutral face expressions either. Having a bunch of ideal shots to build from with perfectly recorded camera data is almost unheard of, although I wish they did it on every production. 3) Definitely good advice, as jawlines and chins can move all over the place. I figured it'd be okay in this shot since I directed him not to move his face or change expressions.
Hello Sir can you help me, i'm fascinating of this awesome tool, but in my case i use 4k pics (3840x2160) i change the level size, all right no problem, when i start to pin after 1-2 min it crashes - everytime, i cant finish 1 head, i have this tool 2 days, and im scary if i use it more my hardware will damage ? crashes are not good at all in this many time. can you see any update to make the performance Epic ? i dont shall or whine, its just awesome, i just hope you fix mabye the support of 4k files or something ? thank you Keentools
The final model and texture will be as good as the photos you use. I think it could be improved if you took the pictures indoors with a diffused light, avoiding both the sun shadows and the model frowning. Also its best if you move the camera for new photos instead of the model moving its head. And finally, a photo from bellow would fix the missing texture in the chin area. This tool is amazing, I've seen a few videos of it and they always blow my mind how accurate and effortlessly it transforms the base mesh, doing it by hand would take several hours, perhaps even days! Great work!
Thanks for this! :) For anyone else new: I'd recommend getting the 1st three points / the camera on each picture sorted first helps before starting any mesh editing. Also for side shots etc that are difficult to get the camera angle right for, the align face tool can help get you started... Also as KeenTools mentions below, corners of the mouth are a good next step after eyes & nose..
Say I made my own head mesh already. It has shape keys and rigs. I don't want to use the one that comes with FaceBuilder addon. Is there a way around where I swap my model head for there's and just tweak my own head mesh and use Facebuilder as normal?
I was researching the nuke plugin a few months back...great to see they're noticing blender now. Next thing blender needs is markerless geometry-based tracking, like syntheyes has, so we can use these heads we make to track with!
@@seankennedy6076 of course, I just realized that KeenTools already has a GeoTracker plugin for Nuke which does exactly that. so maybe that'll be the next thing they port to blender!
Geotracker in blender would be a major game changer at least for me who don't have a pro camera that I know sensor withd and other stuff on my phone camera, geometry track gives the software a larger chunk of pixels and thos makes it a bit easier to track
There is a dedicated lens for taking portraits, 105mm. or position far enough away to cut out distortion and then crop. But you can achieve a better result using Autodesk 123D Catch
I would love to try that, but the images I took of the subject were different sizes - it was from a match. The lens was the same, but the focal lenght was changing and subject was closer or further away in shots. This probably wouldn't work inside this software?
I would guess no. However, you could try cheating the image sizes by bring them all into an image editing program like Photoshop or GIMP and dropping them all into the same size image frame and re-saving them. You may have some modeling misalignments because of the focal length change, but maybe this would at least allow you to use the images you have.
Hi.great Tut. Need help Please, when I add images and press the view button to switch to pin mode, All the images showing upside down?! What am I doing wrong?? please help
Glad you liked it! I've got a Geotracker tutorial coming on in the next couple days, made by the same company! Hope you'll come back and check that out too!
Og seems like you dont really get the laugh marks by the nose, between the cheeks and the mouth section. Or well, only texturally. Is there any way to get those to be a part of the base mesh itself?
Not that I know of, however KeenTools themselves may be able to dig into that further. I wonder if you could place enough pins in to really get into finer details like that? I doubt it.
What would I use this application for? Can you create a tut where you'd use a 3D head for some reason. Is it to replace a stuntman or other person? is it for far off distant shots? why would I need to use this and in what sense. Thanks.
Yeah, I actually have a head tracking tutorial kind of in the works. For some ideas of things you can do with this, check out the matchmving by hand tutorials I did a while ago. openvisualfx.com/2019/10/04/matchmoving-by-hand/ That should give you a basic idea. There's so much you can do!
As for the body, you could do that using traditional methods, modeling a body and then manually projecting the texture on to it, then cleaning it up with paint tools.
I took images of my own face I saved them into a new folder called faces. I click add images in face builder that folder comes out empty. So I click on add scroolled down to empty and then scrolled to images went to the menu on the right side of the interface and opened up the image from there. I do not know why it doesn't work in the face builder
Is it possible to use video as a source instead of image sets? That would solve the problem of finding multiple images that share the same camera settings & resolution
I'm pretty sure that's showing you how much the pin has to drift from it's original position because the add-on cannot warp the face to all the pins at the same time. So it slightly shifts the pin placement.
As far as I can tell, no limits on the beta. Yes, you can save. Nothing really to export, it's just a mesh head that will always be openable in Blender, but sure if you want to export it as alembic or OBJ to some other program, that would work just like any other geometry you make in Blender. The only catch with the plugin is that once the beta testing is over, it becomes a paid plugin.
Anyone know if there's any way to change the face model to a different one? I have a different face obj I wanted to use but I don't know how to change the default head. Hope this makes sense
As far as I know, I don't think it's possible. Facebuilder lets you turn off parts of the geometry you don't want to work with, like the back of the head or the ears, and I imagine that feature would entirely break if users could use their own head models. I'm sure other features would break, too.
Thanks a lot, after I added the image and try to open it , blender directly shutdown, why ? and i need your help to find image collection are compatible with your tutorial for trying and learning... 🌹🙏
I don't know why Blender crashed for you, but I've posted download links to the photos I use in both my Facebuilder tutorials over at www.openvisualfx.com, hope that helps!
Cool add-on. but how do I fix my images loading in sideways? I've played around with this for a while and still have no idea. Maybe I"ll try to rotate them in GIMP and see if that helps. I'm sure there's an easy answer, I just don''t know what that would be.
That's something I've not done very much of. The gang over at CG Cookie have put out a few good ones, though, you should check them out! cgcookie.com/tutorial/modeling-stylized-hair-in-blender ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-WVJyGzPJIeQ.html
Does anyone know if theres a way to preserve the facial expression? I have a pose with someone with their mouth open and I'd love to maintain that pose rather than automatically 'reseting' & closing the mouth once Pin mode is left. Any help would be great, thanks!
It could help you create a specific head shape, which you could then modify to be a bas relief. But I don't think it could create that bas relief style on it's own.
How can i export all the head into a format .obj or .ply with texture already attached to it. I open the file in another program but i don't get the texture attached to it, i only get a head.
Hmmm, not sure, actually, since I haven't exported it from Blender to any other program yet. OBJ isn't working? Usually that also gives you a MTL file which should be the material.
The face looks really good but the whole face is plain. The geometry is really good to. I do not have the blue around the eyes ears mouth and nose. So when I hold the mouse to move the pin it just moves that geometry It does not move the rest of the geometry in place. So my face builder is not working right at all. I think its because of those blue lines. How do get the blue lines on the face.
Are you still having this problem? There's a small checkbox at the bottom of the Wireframe panel in the Facebuilder tools, called "Highlight Head Parts". That should highlight those areas in blue.
Yeah I got it when I name my images I have to remember to press.png files name. Then I named a new folder so its easy to get to. Thank you for getting back to. Now I just need to find some good face images to mess around with.
@@daviddahl8562 I shared the images I used in the tutorial, so please feel free to use those until you can take some of your own! :) Link in the description
@@seankennedy6076 TRYING to get this, I have like 50 pins and there is no smooth way to to move the mesh. Using an image and now it looks like a circus freak.... there MUST be something I am missing. Just can not figure it out. Maybe i should do a screenshot vid of what I am doing.
@@coreyfairbanks7374 Yeah, feel free to send a screenshot of what you're doing to my Contact Me page over at OpenVisualFX.com. I'll take a look as soon as I can.
@@seankennedy6076 I was meaning like imprint the texture on the shapes mesh somehow maybe. Itd be cool if it'd let you sculpt while the texture is showing so you can trace an engraving of the fine details into the mesh instead of trying to detail it by eye maybe.
I swear no matter what tutorial I watch, my menu is always different. I have the newest version of Blender (3.5), but when I generate the new head, there is no Add Images button to click on. What am I doing wrong? It's this way with every damn thing I try to do in Blender. My menu is always different than the one in the video. To say that I'm getting very frustrated is understatement of the year...
Even after 10 trials didnt manage to set the third pin, after the first two seeming to work, at the third one invariably when draging on it everything goes boom, with mesh in one direction and picture in the opposite one. Tryed uninstall, install again a couple of times, to no avail. I just gave up, seems more intelligent thing to do.
Bummer it wasn't working good for you. If you ever want to try again and have problems, do a little screencapture for me and send it through the contact me page at the website, I'd be happy to help if I can.
Huh, wow, I hope they didn't get rid of the trial, that'd be a bummer! You could still manage to do this without plugins, just bring in a generic head model, use soft selection in vertex edit mode, and slightly adjust sections of the face and head to your reference photos. The biggest drawback of not using this tool is that you'd have to realign the model to every new reference image you import, which will get tedious, but is totally possible.
@@seankennedy6076 yeah. I would assume you just cut off the head for the mesh and join the head model. But it seems too easy. I’m sure I’m missing something lol
Everybody says the pictures you use should be from the same source. No, they _must_ have the same focal length and "frame size," otherwise you'll get an ugly error message and it won't work properly. This technology is is interesting, but I don't think it will really hit its stride until it's less finicky about the kinds of photos you use.
Good point about the pictures. And this tech definitely feel more at home in a video/film environment. Creating a face from a single shot of an actor, just using different frames, is extremely fast and easy.
@@seankennedy6076 Maybe I'm doing it wrong? If I only have one photo, it just sets the frame to the size of the picture. How would you change the frame for a single photo, and why would you want to?
@@FreakazoidRobots Oh, I probably wasn't clear enough in my response. At work, we don't use it on a single photo, but on a single scene, a scene that was shot of an actor acting. So we go through the scene, through the various frames, adjusting the model as we go. So we may actually end up using more than 7 or 8 images. But because all the frames of the scene are from the same camera and lens, it's usually not an issue to deal with the different camera and lens problem.
@@seankennedy6076 Ah, it threw me off when you said "actor." I'm not a member of a studio. I'm just some guy trying to learn the technology. I've been using public photos of Gal Gadot, because what's the fun of using a 3D modeling program if you can't use it to create a celebrity? Obviously, under those circumstances it's impossible to have that much control over things. I was able to get a reasonable likeness from a single photo as far as the mesh goes, but it's been an unbelievable bear trying to fix the texture. It's photo realistic, but that's a problem because shadows are baked in. That seems like a problem if you want to animate the model in the future. I spent all day today fruitlessly trying to remove the shadows using light curve masks in Affinity and I'm starting to wonder if it would be easier to just create a new texture from scratch. That'd be a shame, because it seems like photo realistic textures of a specific person would be _hard_ to create from scratch and I was hoping the program could save me some time.
@@FreakazoidRobots Yeah, painting the specular highlights and shadows out of a texture image is a huge pain in the butt. Definitely been there before. What I like to do is duplicate the base image, adjust the levels until the shadow areas look as close as possible to the evenly lit areas of the base image, then put a black layer mask on that top layer and paint it back in to the shadow areas using a 5% or 10% brush. Gives you really good control!
Are you asking how to make textures for the back of the head? Just be sure to use some pictures of the back of the head (and sides) while creating the face, the textures should get created automatically like the rest of them.
Just because it's paid, I'll never use it as a freelancer. Blender is big because volunteers made it so. It's disrespectful to exploit the community's efforts by making paid add-ons on an Opensource platform. If I need to do the effects mentioned, I'll build my own tool or rig to accomplish the same thing. Some time back, I was ripped-off almost $200 by a professional add-on... not making that mistake again.
Wow, that sucks about getting ripped off on the $200 add-on! I disagree with you about it being disrespectful to sell tools for an open source program, but everyone is certainly entitled to their opinion. :) And I even mention in the tutorial that this is still a pretty simple thing to do without a dedicated add-on. Hopefully I at least made you aware of it's capabilities!
@@seankennedy6076 well, if you had said that... i would've been left with less to make a joke about. funny thing is, i made that comment the first time you said it, but you actually said it a second time... except even worse... because you said it's a "great" round number! i felt i had already made my point, though... no need to rub it in. hehe
Because it is moving every point with every picture it is a constant fight against the head reference mesh. Like this it's not possible to get any accurate face out of this even with 100 pictures. This tool is such a low quality. Charging money for it is beyond belife. Worst tool ever made, just BS big time.